Context-dependency in carnivore co- occurrence across a multiuse conservation landscape

Page created by Christopher Gregory
 
CONTINUE READING
Context-dependency in carnivore co- occurrence across a multiuse conservation landscape
20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Received: 11 February 2022 | Revised: 1 August 2022 | Accepted: 3 August 2022

DOI: 10.1002/ece3.9239

RESEARCH ARTICLE

Context-­dependency in carnivore co-­occurrence across a multi-­
use conservation landscape

Gonçalo Curveira-­Santos1 | Laura Gigliotti2 | Chris Sutherland3 | Daniela Rato1 |
Margarida Santos-­Reis1 | Lourens H. Swanepoel4,5

1
 Centre for Ecology, Evolution and
Environmental Changes (cE3c), Faculdade Abstract
de Ciências, Universidade de Lisboa,
 Carnivore intraguild dynamics depend on a complex interplay of environmental affini-
Lisbon, Portugal
2
 Department of Environmental Science,
 ties and interspecific interactions. Context-­dependency is commonly expected with
Policy, and Management, University of varying suites of interacting species and environmental conditions but seldom em-
California Berkeley, Berkeley, CA, USA
3
 pirically described. In South Africa, decentralized approaches to conservation and the
 Centre for Research into Ecological and
Environmental Modelling, University of St resulting multi-­tenure conservation landscapes have markedly altered the environ-
Andrews, St Andrews, UK mental stage that shapes the structure of local carnivore assemblages. We explored
4
 Department of Zoology, School of
 assemblage-­wide patterns of carnivore spatial (residual occupancy probability) and
Mathematical & Natural Sciences,
University of Venda, Thohoyandou, South temporal (diel activity overlap) co-­occurrence across three adjacent wildlife-­oriented
Africa
5
 management contexts—­a provincial protected area, a private ecotourism reserve, and
 African Institute for Conservation
Ecology, Levubu, South Africa commercial game ranches. We found that carnivores were generally distributed in-
 dependently across space, but existing spatial dependencies were context-­specific.
Correspondence
Gonçalo Curveira-­Santos, Centre for Spatial overlap was most common in the protected area, where species occur at
Ecology, Evolution and Environmental higher relative abundances, and in game ranches, where predator persecution pre-
Changes (cE3c), Faculdade de Ciências,
Universidade de Lisboa, Lisbon, Portugal. sumably narrows the scope for spatial asymmetries. In the private reserve, spatial
Email: gcurveirasantos@gmail.com co-­occurrence patterns were more heterogeneous but did not follow a dominance
Funding information hierarchy associated with higher apex predator densities. Pair-­specific variability sug-
African Institute for Conservation gests that subordinate carnivores may alternate between pre-­emptive behavioral
Ecology; Fundação para a Ciência e
a Tecnologia, Grant/Award Number: strategies and fine-­scale co-­occurrence with dominant competitors. Consistency in
PD/BD/114037/2015 and UID/ species-­pairs diel activity asynchrony suggested that temporal overlap patterns in our
BIA/00329/2019; National Geographic
Society, Grant/Award Number: EC-­ study areas mostly depend on species' endogenous clock rather than the local con-
314R-­18; South African Agency for text. Collectively, our research highlights the complexity and context-­dependency of
Science and Technology Advancement,
Grant/Award Number: 107099 and guild-­level implications of current management and conservation paradigms; specifi-
115040; Wild Tomorrow Fund cally, the unheeded potential for interventions to influence the local network of carni-
 vore interactions with unknown population-­level and cascading effects.

 KEYWORDS
 camera trap, conservation management, co-­occupancy, interspecific interactions, temporal
 overlap

 TA X O N O M Y C L A S S I F I C AT I O N
 Applied ecology; Community ecology; Landscape planning; Spatial ecology; Zoology

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2022 The Authors. Ecology and Evolution published by John Wiley & Sons Ltd.

Ecology and Evolution. 2022;12:e9239.  www.ecolevol.org | 1 of 17
https://doi.org/10.1002/ece3.9239
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 of 17 CURVEIRA-­SANTOS et al.

1 | I NTRO D U C TI O N effects by, for instance, reducing risk-­free and undisturbed space,
 inducing nocturnality, providing reduced or surplus resources (e.g.,
How species' resource use affinities and interspecific interactions Curveira-­Santos et al., 2017; Mills & Harris, 2020). Hence, anthro-
act to structure animal communities is a central tenet of coexistence pogenic influence may artificially narrow or dilute spatiotemporal
theory (Macarthur & Levins, 1967; Schoener, 1974). Research on scopes for carnivore interactions, thus altering the landscape of co-
spatial and temporal species' co-­occurrence patterns (i.e., relative existence (Oriol-­cotterill et al., 2015). Moreover, anthropic influence
spatial and diel activity distributions) within mammalian carnivore on apex predators is particularly accentuated and well-­described,
guilds has assumed particular relevance (e.g., Davis et al., 2018). ranging from harmful interventions (e.g., direct persecution; Treves
This is largely due to carnivores' propensity for agonistic interac- & Karanth, 2003) to directed conservation initiatives (Mossaz
tions (Caro & Stoner, 2003; Donadio & Buskirk, 2006; Ritchie & et al., 2015; Sergio et al., 2008), influencing these species' ability
Johnson, 2009) and their influence over ecosystem processes and to regulate guild and ecosystem dynamics (Dorresteijn et al., 2015;
functioning (Estes et al., 2011; Prugh et al., 2009; Ripple et al., 2014). Haswell et al., 2015; Kuijper et al., 2016). Carnivore communities of
Importantly, ongoing global declines of apex predators and shifts to- similar composition may thus exhibit fundamentally different spatio-
ward mesopredator-­dominated systems (Hoeks et al., 2020) have temporal structures across ecological contexts and, subsequently,
motivated increasing calls for a more comprehensive understanding varying interaction dynamics.
of predator community structure and its inclusion in conservation South Africa's diverse carnivore assemblages and the complexity
and restoration plans (Jachowski et al., 2020; Ritchie et al., 2012; of the highly variable local conservation landscape are a pertinent
Ritchie & Johnson, 2009; Svenning et al., 2016). model system to explore context-­specific carnivore co-­occurrence
 Fine-­scale coexistence of sympatric carnivores species is me- patterns. Intraguild killing and interference exploitative competi-
diated by a complex interplay of interspecific interactions beyond tion are pervasive among African carnivores (Caro & Stoner, 2003).
individual resource preferences and species' fundamental niches Moreover, empirical evidence supports suppressive effects by the
(Rosenzweig, 1966). Behavioral adjustments associated with inter- dominant apex predator, the African lion (Panthera leo), and other
specific interactions include changes in space use and circadian ac- large carnivores, while the wide range of carnivore species' body
tivity, to avoid confrontation with dominant species and/or partition sizes presupposes potential suppression-­based cascades (Levi &
the use of common resources (Karanth et al., 2017; Mills et al., 2019; Wilmers, 2012). Carnivore sympatry in South African landscapes is
Monterroso et al., 2014, 2020; Robinson et al., 2014). Outcomes of thus expected to arise via a complex interplay of behavioral mech-
antagonistic interactions may induce suppression-­driven cascades anisms and co-­
 occurrence patterns (Durant, 1998; Hayward &
whereby apex predators limit large-­bodied mesopredators, indirectly Slotow, 2009; Mills et al., 2019; Ramesh et al., 2017).
benefiting smaller carnivores (Levi & Wilmers, 2012). Conversely, Notably, however, widespread human encroachment, wild-
facilitative interactions, such as carrion provisioning by large car- life fencing (Packer et al., 2013), and decentralized approaches to
nivore hunts, may promote carnivore co-­occurrence and even en- conservation (Pitman et al., 2017) have markedly altered the stage
hance suppression at the population level (Prugh & Sivy, 2020; Sivy upon which carnivore interactions take place. Private solutions to
et al., 2017). When interspecific aggression and competition are not natural resource management have increasingly complemented
the primary drivers, co-­occurrence in space and time can result from the formal network of protected areas following changes in wild-
trait similarities and common environmental and resource affinities life ownership rights, which prompted a large-­scale land-­use shift
or stressors (i.e., habitat filtering; Rich et al., 2017). The interplay of from livestock farming to game ranching (ecotourism and/or hunt-
co-­occurrence patterns becomes particularly diffuse in species-­rich ing; Pitman et al., 2017), giving rise to intricate multi-­tenure con-
carnivore assemblages, where heterogeneity in species' behavior, servation landscapes (Curveira-­Santos, Sutherland, Santos-­Reis, &
morphology, and phylogeny influences the nature and strength of Swanepoel, 2021a; Di Minin et al., 2013). While large carnivores
interspecific interactions at local scales. have been widely reintroduced into small and fenced reserves and
 A growing body of literature has helped elucidate the complexity often maintained at high densities (Mossaz et al., 2015), human pres-
of carnivore community structure and intraguild interactions. Less at- ence and persecution of free-­ranging species on game ranchland
tention, however, has been given to underlying context-­dependency, is common practice (Lindsey et al., 2008). Changes in guild com-
which is generally assumed but often poorly described (Bar-­Massada position are accompanied by varying levels of human disturbance
& Belmaker, 2017; Chamberlain et al., 2014). Differences in commu- and its influence on resource availability across multiple land uses.
nity composition and abundance of local carnivore species, as well as Together, these can deeply shape the structure of carnivore assem-
diversity, availability, or spatial structuring of resources, can underly blages (Curveira-­Santos, Sutherland, Tenan, et al., 2021b; Schuette
context-­specific carnivore spatiotemporal dependencies (Karanth et al., 2013) and potentially alter the underlying network of spe-
et al., 2017). Human-­caused disturbances are often the dominant cies interactions and community regulation pathways (Dorresteijn
driver in forming ecological context, by modifying the landscapes et al., 2015). Nonetheless, context-­dependent changes in carnivore
and communities that predators interact with and within, albeit with assemblages have received little attention in carnivore-­rich regions
varying effects in direction and magnitude (reviewed in Sévêque of southern Africa, where context-­dependency is likely to have pro-
et al., 2020). Human influence often has broad and heterogeneous found effects.
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CURVEIRA-­SANTOS et al. 3 of 17

 Here, we link heterogeneity in carnivore co-­occurrence patterns the production of wild ungulate species, and occasionally domestic
to variation in conservation management models implemented in cattle, with large expanses of natural habitat and low human densi-
South Africa. We evaluate spatial (residual occupancy correlation) ties. Leopards (Panthera pardus) and spotted hyenas (Crocuta Crocuta)
and temporal (diel activity overlap) carnivore co-­occurrence pat- are the only large carnivores that are observed in GR, where both
terns across three adjacent wildlife-­oriented management contexts, experience widespread persecution (Balme et al., 2010). The species
spanning a 109-­
 year-­
 old provincial protected area (“conserva- composition of medium-­ (5–­20 kg) and small-­sized ( 0 . Within the indicator
 ( )

reserves hold a near-­complete suite of large carnivores, although function, I(.), the variance of u ij is constrained by covariate effects
African wild dogs (Lycaon pictus) only occasionally visit the PR. The and by a set of T latent variables lj = lj1 , … ,ljT and corresponding
 ( )

mosaic of commercial game ranches (GR) to the south is dedicated to species-­specific latent variable coefficients i = i1 , … , iT . The
 ( )
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 of 17 CURVEIRA-­SANTOS et al.

 F I G U R E 1 Camera-­trapping surveys
 across three adjacent wildlife-­oriented
 management contexts (provincial
 protected area, private ecotourism
 reserve, and commercial game ranches)
 in the Maputaland region of northern
 KwaZulu-­Natal, South Africa

latent variables are specified as random variables from a standard (TREE). We also considered site-­level (camera) covariates influencing
normal distribution and represent unmeasured site-­level variation variation in detection probability; specifically, the average width of
that is attributable to species spatial dependencies. The variance the trail structure targeted (TRAIL_W), and mean enhanced vegeta-
of the residuals εij accounts for the variance absorbed by the la- tion index values (MODIS EVI datasets: https://lpdaac.usgs.gov/) for
tent variables and is derived from adjusted variance σi2 values for the survey period as a proxy the vegetation density (VEG_D) in the
each species i. For T latent variables and n species, the full species immediate vicinity (30 m) of each site. Prior to analysis, we standard-
correlation matrix R is derived from the correlation in the latent ized all area-­specific covariates to have a mean of 0 and standard
variables as R = T + diag 21 , 22 , … , 2n . deviation of 1.
 ( )

 We were interested in the residual correlation in the occupancy In order to parse out context-­specific residual occupancy cor-
probability that cannot be explained by the environmental covariates relation patterns, we fitted a single global JSDM for each target
in the model, that is, after accounting for species-­specific baseline area, with the following shared formulation:
environmental preferences. Therefore, we modeled species-­specific (i) Occupancy
site occupancy probability as a function of covariates shown to in- ( )
 Zij = I uij > 0
fluence species' occupancy patterns in this dataset (see Curveira-­
Santos, Sutherland, Santos-­Reis, & Swanepoel, 2021a for details),
 uij = 0,i + 1,i TREEj + lj j + ij
namely remote-­sensed tree cover estimates within a 500 m radius
buffer around each camera station (MODIS vegetation continuous
 T
fields; DiMiceli et al., 2011) as a measure of the spectrum of vegeta-
 ∑
 2i = 1 − 2it
tion structure ranging from open grasslands to woodland savannas t=1
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CURVEIRA-­SANTOS et al. 5 of 17

 ij ∼ Normal 0, 2i co-­occurrence estimates as a function of the species-­pair traits.
 ( )

 For spatial co-­occurrence patterns, we fit generalized linear mod-
 (ii) Detection els with the pairwise residual occupancy correlation as the response
 ( ) variable assuming Gaussian errors and an identity link to accommo-
 yij ∼ Binomial kj ,zij x pij
 date the [−1,1] distribution of this response variable. For temporal
 ( ) co-­occurrence, we implemented beta regression models with the
 logit pi,j = 0,i + 1,i TRAIL _ Wj + 2,i VEG _ Dj
 temporal activity overlap coefficient as the response variable given
 the [0,1] distribution of these estimates (Douma & Weedon, 2019).
 The species-­specific regression coefficients 1:2,i and 1:2,i (say ) For each response, we created an a priori global model based on
are treated as species-­specific random effects from a community-­ the additive effect of the dominance hierarchy within the carnivore
level distribution: guild (i.e., proxy for top-­down pressure) and pairwise measures of
 ecological similarity. To formally identify and describe how the na-
 ture and strength of each covariate effect varies across management
 ( )
 i ∼ Normal , 
 contexts, we modeled all covariate effects as interacting with the
We fitted the JSDMs with 7, 6, and 5 latent variables for PA, PR, and three-­level “area” covariate (i.e., PA, PR, and GR).
GR, respectively, corresponding to about n/2 latent variables neces- Dominance hierarchy covariates included the different combi-
sary to approximate the residual correlation matrix when there are nations of species' size-­b ased hierarchy ranks (e.g., “Apex|Large,”
n species in a community (Tobler et al., 2019). We implemented our and “Large|Medium”) coded as binary variables (Appendix B:
models in the BUGS language using the JAGS software (Plummer, Table B1). Lions were ranked as the apex predator, non-­apex spe-
2003) through R. For each co-­occurrence model, we generated three cies >20 kg as large carnivores, species 5–­20 kg as medium-­sized
MCMC chains with 30,000 iterations after a 10,000 iteration burn-­in carnivores, and species 1 h after the last detection of the same spe- categories divided by the total number of unique food categories
cies, converted to solar time to facilitate ecological interpretation. used by the pair; from Caro & Stoner, 2003), whether a pair in-
To describe the general activity period of each species, we used cluded two species of the same hierarchy rank (proxy for lateral
the function “modal.region.circular” to calculate the 95% activity competition), and whether a pair included species from the same
isopleth. We used the function “getBandWidth” to calculate the family (phylogeny effect, proxy for relatedness). Additionally, we
best smoothing parameter (κ) for each species and maintained the modeled the effect of the body-­mass ratio of pairs (including a
highest value when comparing activity patterns. Finally, using the quadratic relationship) to represent the suggested prevalence of
function “totalvariation,” we calculated pairwise conditional activ- agonistic carnivore interactions at intermediate body-­size differ-
ity overlap coefficients as a measure of temporal co-­o ccurrence; ences (body-­mass ratios between 2 and 5.4) when the species
ranging from 0, for perfect activity dissimilarity, to 1, for full inter- are similar enough in size to compete for same prey but different
section of activity periods (95% isopleths). enough so that the larger size of the aggressor entails a low risk
 of injury (Donadio & Buskirk, 2006). This variable was highly cor-
 related with diet overlap (r = 0.70), and to prevent collinearity is-
2.4 | Drivers of co-­occurrence sues (Dormann et al., 2013), we retained only the body-­mass ratio
 in the global models since it led to an higher R 2 for both the spatial
To investigate drivers of spatial and temporal co-­occurrence and and temporal dimensions. Finally, we considered the effects of the
how these vary across management contexts, we modeled our species spatial co-­o ccurrence on temporal overlap and temporal
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 of 17 CURVEIRA-­SANTOS et al.

overlap on spatial co-­o ccurrence to evaluate complementarity in co-­occurrence between strictly nocturnal and diurnal species (0.00
niche partitioning patterns (Schoener, 1974). between slender [Herpestes sanguineus] and white-­t ailed mongooses
 in both the PA and GR) to nearly complete activity overlap (0.97 be-
 tween leopard and serval [Leptailurus serval] in PA), thus indicating
3 | R E S U LT S an intricate temporal co-­occurrence structure at the assemblage
 level (Figure 3b). Nonetheless, moderate-­
 to-­
 high overlap values
Our dataset included a total of 4895 independent records of our (>0.5) were common in all three areas (69% in PA, 65% in PR, and
13 target wild carnivore species over 19,739 effective trap-­days 57% in GR) due to a high number of predominantly nocturnal species
(Appendix A: Table A2). We estimated spatial (residual correlation in all the assemblages. While the distribution of activity overlap val-
in occupancy probability) and temporal (diel activity overlap) co-­ ues was similar between the protected area and the private reserve,
occurrence for a total of 161 species pairs: 78 in PA, 55 in PR, and the pattern in the game ranches was bimodal, that is, either low or
28 in GR (Figure 2). high overlap. This was mostly attributable to the absence of several
 species with cathemeral activity patterns (e.g., lion and serval) rather
 than accentuated species-­specific changes in activity (see below).
3.1 | Spatial co-­occurrence

The mean residual correlation in occupancy across all species 3.3 | Context-­dependency in co-­occurrence
pair and area combinations was 0.06 ± 0.15 (mean ± SD), indicat- patterns
ing that, at the assemblage level, there was neutral or independ-
ent spatial co-­o ccurrence. Area-­specific means exhibited a similar Considering only species pairs present in more than one area
pattern (PA = 0.06 ± 0.12, PR = 0.03 ± 0.16, and GR = 0.08 ± 0.11), (n = 55), context-­dependency in pairwise spatial dependencies was
albeit with large variation across species pairs (Figure 3a). Pairwise frequent whereas temporal co-­
 occurrence patterns were largely
residual occupancy correlation values ranged from 0.46 (95% consistent across areas (Figures 2 and 4; Appendix C: Figures C1 and
Bayesian credible intervals, 0.27–­0 .64), between the spotted hy- C2). Although weak residual occupancy correlation strength was the
aena and the white-­t ailed mongoose (Ichneumia albicauda) in the norm, nearly half (55%) of species pairs exhibited contrasting sig-
private reserve, to −0.36 (−0.52 to −0.05), between the lion and nals in residual occupancy correlations across areas, that is, context-­
the large-­spotted genet (Genetta maculata) also in the private re- dependency. Species with high probability of spatial avoidance in
serve (Figure 2a; Appendix C: Figure C1). Positive spatial depend- one context (e.g., lion and large-­spotted genet in PA), specifically,
encies were predominant in both the protected area (67%) and the rarely did so in the others. Consistency in positive spatial dependen-
game ranches (68%), while positive and negative residual correla- cies across contexts was more common; however, seldom pairs ex-
tion values were more evenly distributed in the private reserve hibited correlation values that strongly departed from a hypothesis
(56% vs. 44%, respectively, Figure 3a). However, strong evidence of independence (>0.9 probability) in different contexts. Conversely,
for non-­independent spatial co-­o ccurrence patterns (>0.9 prob- species' temporal co-­occurrence patterns were similar between con-
ability of a different than zero correlation) was only observed in 14 texts, with 75% of pair-­specific diel activity overlap differences of
out of 161 pair-­by-­area combinations; 12 of which were positive ≤0.1 across areas.
(five in PA, five in PR, and three in GR) and only two were negative
(one in PA and PR each). Notably, the strength of negative spa-
tial co-­o ccurrence signals in game ranches was very low (>−0.1), 3.4 | Drivers of co-­occurrence
with no pairs exhibiting a >0.7 probability of a spatial avoidance
pattern. The main drivers of carnivore spatial co-­
 occurrence differed
 among management contexts (Table 1a). In the protected area,
 “Apex|Small” dominance hierarchy pairs, that is, pairs with the lion
3.2 | Temporal co-­occurrence and a small-­sized carnivore, exhibited lower spatial co-­occurrence
 relative to all other pairs. In the private reserve, spatial co-­
Pairwise temporal activity overlap coefficients spanned a wide occurrence was lower for pairs from the same taxonomic family,
range of values (Figure 2b; Appendix C: Figure C2), from no temporal but increased for pairs with higher diel activity overlap (i.e., pairs

F I G U R E 2 Schematic depictions of pairwise (a) spatial and (b) temporal co-­occurrence estimates within South African carnivore
assemblages, across three adjacent management contexts (PA—­provincial protected area; PR—­private game reserve; GR—­commercial
game ranches). Body size in the carnivore assemblage decreases clockwise. Spatial co-­occurrence patterns were estimated as pairwise
residual occupancy correlation values from a hierarchical Bayesian joint species distribution model, and temporal co-­occurrence expressed
as pairwise coefficients of diel activity overlap from non-­parametric, circular kernel density functions (see Methods). Line widths are
proportional to the estimated co-­occurrence values.
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CURVEIRA-­SANTOS et al. 7 of 17

 (a) (b)

with high temporal overlap also exhibited high spatial overlap) and of the pairs' body-­mass ratio. No clear associations were found to
of intermediate body-­mass differences (ratios ~2), that is, the re- explain variability in species' spatial co-­occurrence in the game
sidual occupancy correlation varied by a concave quadratic effect ranches.
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 of 17 CURVEIRA-­SANTOS et al.

(a) (b)

F I G U R E 3 Assemblage-­level (a) spatial (residual occupancy correlation) and (b) temporal (diel activity overlap) carnivore co-­occurrence
patterns in each of the three management contexts (PA—­provincial protected area; PR—­private game reserve; GR—­commercial game
ranches). Vertical lines mark area-­specific medians (solid line) and 33 and 66% quantiles (dashed lines).

 The global model for the drivers of carnivore temporal co-­ 4 | DISCUSSION
occurrence (Table 1b) indicated that, in all areas, pairings that in-
cluded two small-­sized species exhibited less temporal overlap than Our study provides important insights into context-­dependency of
other size pairings. Additionally, only in the private reserve, spe- carnivore spatial and temporal co-­occurrence across a South African
cies with high spatial overlap also exhibited high temporal activity multi-­
 tenure conservation landscape. We found that carnivores
overlap. generally distribute independently across space. Clear pairwise
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CURVEIRA-­SANTOS et al. 9 of 17

(a)

(b)

F I G U R E 4 Context-­dependency in (a) spatial (residual occupancy correlation) and (b) temporal (diel activity overlap) co-­occurrence
patterns for species pairs present in two or more areas (n = 55). PA, provincial protected area; PR, private game reserve; GR, commercial
game ranches. For the spatial dimension, points in the top-­right and bottom-­left quadrants indicate a consistent co-­occurrence signal, while
top-­left and bottom-­right quadrants indicate context-­dependent spatial dependencies. The color and shape of the points represents the
proportion of the posterior distribution of the residual occupancy estimate with the same sign as the mean, that is, the probability for a
different than zero correlation. For the temporal dimension, consistency in diel activity overlapped is expressed by proximity to the diagonal
bar. Credible and confidence intervals were omitted for visual clarity.

TA B L E 1 Beta coefficient estimates (±SE) for species-­pair-­specific covariates representing the hypothesized drivers of carnivore (a)
spatial and (b) temporal co-­occurrence

 (a) Spatial co-­occurrence (b) Temporal co-­occurrence

 Covariate PA PR GR PA PR GR

 Dominance hierarchy Apex|Small −0.18 ± 0.08 0.13 ± 0.08 -­ -­ -­ -­
 Large|Medium -­ -­ -­ 0.58 ± 0.34 0.54 ± 0.46 0.98 ± 0.57
 Same rank Large −0.07 ± 0.06 0.14 ± 0.09 0.21 ± 0.16 0.17 ± 0.46 0.20 ± 0.65 −0.91 ± 1.23
 Medium −0.09 ± 0.10 −0.08 ± 0.16 0.02 ± 0.17 0.43 ± 0.74 1.23 ± 1.14 0.13 ± 1.27
 Small −0.02 ± 0.06 0.12 ± 0.07 −0.02 ± 0.11 −1.73 ± 0.42 −1.15 ± 0.50 −1.98 ± 0.75
 Same family 0.07 ± 0.04 −0.18 ± 0.06 −0.02 ± 0.11 0.46 ± 0.33 0.30 ± 0.47 −0.27 ± 0.84
 Body-­mass ratio −0.10 ± 0.05 0.14 ± 0.06 0.03 ± 0.10 −0.15 ± 0.37 0.11 ± 0.45 −0.78 ± 0.71
 Body-­mass ratio2 0.02 ± 0.01 −0.03 ± 0.01 −0.01 ± 0.02 −0.01 ± 0.06 −0.04 ± 0.07 0.07 ± 0.13
 Temporal co-­occurrence −0.08 ± 0.06 0.22 ± 0.08 0.06 ± 0.09 -­ -­ -­
 Spatial co-­occurrence -­ -­ -­ −1.04 ± 0.89 2.10 ± 0.84 1.70 ± 1.71
 Global model R 2 0.25 0.39

Note: Well-­supported coefficients (p < .05) are bolded.
Abbreviations: PA, provincial protected area; PR, private game reserve; GR, commercial game ranches.
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 of 17 CURVEIRA-­SANTOS et al.

spatial dependencies were rare in all areas and seldom consistent Despite the predominance of independent or overlapping spatial
across management and conservation models. Yet, as hypothesized, patterns, previous studies have suggested that, albeit rare, spatial
a higher overall degree of spatial overlap was observed in the pro- avoidance is most common for pairs including large-­
 bodied car-
tected area, where species generally occur at higher relative abun- nivores (Davis et al., 2018), able to exert strong suppression over
dances, and in game ranches, where human disturbance presumably subordinate species and induce avoidance behaviors (Donadio &
narrows the scope for spatial asymmetries. In the private reserve, Buskirk, 2006; Ritchie & Johnson, 2009). We were particularly inter-
spatial co-­occurrence patterns were more heterogeneous but did ested in how the spatial structuring of the competitive environment
not support the hypothesis of a well-­marked dominance hierarchy may uphold coexistence between competition-­
 depressed meso-
associated with higher apex predator densities. In contrast, species-­ predators and the densely distributed apex predator population in
pairs diel activity overlap was widespread and stable across areas, the private reserve. The few depressed residual occupancy correla-
suggesting that, in our study areas, temporal asynchrony is less af- tion values observed included variable body-­size pairings and domi-
fected by the local context but regulated by each species' endoge- nance hierarchies. However, particularly in the protected area, small
nous clock and adaptations to long-­term coexistence with dominant carnivore species tended to spatially avoid lions. This is in line the
predators. Our results suggest that carnivore interspecific interac- with recently uncovered susceptibility of small African carnivores
tions may have limited expression in species co-­occurrence patterns, to lion suppression (Curveira-­Santos et al., 2022; Curveira-­Santos,
but existing spatial dependencies may reflect varying outcomes of Sutherland, Tenan, et al., 2021b). Unlike larger subordinate species,
context-­specific interactions. better equipped to escape and mitigate dangerous encounters while
 Irrespective of the context, the tendency for weak residual co-­occurring with the apex predator, for small carnivores spatial
spatial dependencies among species pairs indicates that carni- avoidance of lions may be a better strategy (Wirsing et al., 2010).
vores were generally distributed independently over the landscape. Collectively, however, our results reinforce the notion that infor-
But note that partitioning data by area for insights on context-­ mation on species spatial co-­occurrence, while useful to character-
dependency likely promotes the underestimation of co-­occurrence ize the setting of potential ecological interactions, may be a poor
strength (Tobler et al., 2019). Among observable spatial dependen- proxy for the actual signal and strength of interactions (Blanchet
cies, we found that carnivore species were more likely to overlap et al., 2020). Detailed behavioral investigations and, importantly,
spatially rather than avoid each other. Such observations corrobo- an increased understanding of carnivore interactions on population
rate global-­scale patterns of carnivore co-­occurrence described by demography remain central to unraveling potential suppression pat-
Davis et al. (2018) and the general assertion that the relative distri- terns (Miller et al., 2018).
bution of sympatric carnivore species is mostly driven by species-­ Temporal partitioning is acknowledged as an important struc-
specific preferences and resource use affinities. Other studies have turing force in carnivore assemblages, facilitating species coexis-
found that sympatric carnivores with similar ecological traits pre- tence (Di Bitetti et al., 2010; Hayward & Slotow, 2009), particularly
dominantly select for the same sites (Davis et al., 2018; Jonathan with increasing assemblage complexity (Monterroso et al., 2014). In
Davies et al., 2007; Monterroso et al., 2020). In our assessment of our multi-­carnivore system, this was evidenced by the full range of
spatial dependencies, we attempted to account for environmental pairwise activity overlap values observed. Diel activity asynchrony
filtering using a broad proxy of habitat structure (i.e., tree cover). was well-­marked between the diurnal species (slender and banded
However, this likely fails to fully capture other, more fine-­scale and [Mungos mungo] mongooses) and remaining carnivores with reduced
resource-­relevant habitat selection factors, such as prey availability, or partial daytime activity (most large-­ and medium-­sized species),
which may underly local scale intraguild sympatry and heterogene- especially sympatric smaller carnivores with predominant nocturnal
ity of the competitive environment (Amarasekare, 2003). Moreover, habits (genet, white-­tailed mongoose and stripped polecat). This
under a common stressor, species may aggregate in spatial refugia corroborates the general understanding of these species' ecology
(Farris et al., 2017; Sogbohossou et al., 2018). In the game ranches, (Hunter, 2018). Among larger species, varying degrees of diurnal
where larger and putative problem species (hyaena and leopard) and crepuscular activity, coupled with frequent asynchronous activ-
experience persecution (Balme et al., 2010; Pitman et al., 2017), ity peaks, often resulted in moderate overlap values. Staple prey of
these carnivores overlapped spatially. In such areas, the influence of large African carnivores are available throughout the day in African
human disturbance on these species may incur fitness costs through savanna ecosystems, and competition avoidance has been proposed
increased competition in shared, low-­risk sites (Sévêque et al., 2020). as the primary cause of temporal partitioning among these carnivore
While the reduced level of spatial asymmetries seemingly contra- species (Hayward & Slotow, 2009).
dicts the “competitive exclusion” and “limiting similarity” principles Remarkably, we found that temporal overlap remained similar
(Macarthur & Levins, 1967), our coarse-­scale analysis of spatial de- across contexts, despite changes in guild composition, species rel-
pendencies precluded exploration of fine-­scale and spatiotempo- ative abundance, and other extrinsic factors. Previous studies have
rally explicit avoidance and resource partitioning mechanisms. Many suggested plasticity in circadian activity patterns of carnivores may
subordinate carnivores minimize the risk of encountering dominant enable coexistence across different ecological contexts (Monterroso
competitors in resource-­rich sites by fine-­scale avoidance behaviors et al., 2014). However, our results suggest that, in our study areas,
and constrained predation strategies (Ramesh et al., 2017). interspecific temporal partitioning among carnivores may be
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CURVEIRA-­SANTOS et al. 11 of 17

maintained within species' own endogenous boundaries, which have spatial segregation more frequent in the private reserve. Compared
evolved under long-­term coexistence with dominant sympatric pred- with the private reserve, the scope for pre-­emptive spatial avoid-
ators. Importantly, low human encroachment in all our focal areas ance may be reduced because of widespread distribution of dom-
likely does not translate into enough pressure to induce increased inant species in the protected area or limited low-­risk spaces in
nocturnality, as observed in other systems (Gaynor et al., 2018). It the game ranches. Detailed behavioral and demographic research
is possible, however, that our area-­level measure of activity overlap is needed to ascertain whether potentially increased carnivore en-
fails to capture micro-­adjustments in activity peaks across contexts. counter rates may enhance competitive and suppression pathways
 Interspecific interactions are multidimensional. Niche parti- or if spatial avoidance in the private reserve is itself the outcome of
tioning theory suggests that in order to reduce competitive stress amplified competitive stress. Nevertheless, our results suggest that
and facilitate coexistence, species converging along one niche understanding nonstationary properties of carnivore interactions
axis, here space or time, segregate in complementary dimensions may be important to avoid erroneous extrapolation to local policies
(Schoener, 1974). However, in both the protected area and game and practices (Rollinson et al., 2021).
ranches, we found that species dependencies across the two studied With our study, we provide novel insights into the potential
dimensions were largely unrelated. Notably, in the private reserve, of alternative management and conservation models to influence
co-­occurrence in space and time was the main underlying assem- community-­wide ecological patterns and processes, specifically, the
blage structure. Again, if environmental and resource affinities, and context-­dependency of spatial and temporal associations of African
not intraguild interactions and competition, are the primary driver of carnivore species. Our empirical exploration of co-­occurrence pat-
carnivore co-­occurrence, such patterns may simply translate to com- terns emphasizes three important aspects underlying carnivore
mon strategies by which groups of sympatric carnivores exploit their community research: the value of multispecies assessments (Heim
environment in each context. Such reasoning reinforces the need to et al., 2019), the importance of predator behavior and interspe-
consider the potential role of fine-­scale behavioral adjustments for cific interactions (Ritchie & Johnson, 2009), and the prevalence
a sound understanding of the mechanisms regulating co-­occurrence of context-­
 dependency in ecological interactions (Chamberlain
patterns (Cusack et al., 2016; Vanak et al., 2013). Importantly, such et al., 2014). Understanding such aspects and how they relate to
spatial and temporal relationships are theoretically underpinned by management interventions, particularly under southern Africa's
the degree of trophic niche sharing (the third and missing fundamen- decentralized conservation approaches and predator-­oriented man-
tal niche axis, Schoener, 1974) and varying prey preferences and agement paradigms (Caro, 2015; Pitman et al., 2017), is of interest
predatory strategies (Hayward & Kerley, 2008). to carnivore conservation efforts and for managing and protecting
 Prevalent patterns of independency in species spatial distri- food webs (Estes et al., 2011; Ritchie et al., 2012). However, our re-
bution and consistent temporal activity overlap across areas did search also exemplifies the challenge of studying multiple species
not fully support the hypothesized relationship between the local and environmental contexts, when resulting patterns are the likely
management context and assemblage-­level signals in species co-­ outcome of a complex web of spatially structured intraguild rela-
occurrence. However, we observed that well-­marked pairwise spa- tionships that may mask individual-­species associations, themselves
tial dependencies were idiosyncratic, with same species occurring ruled by inconspicuous idiocracies of species behavior.
independently or with opposing signal in one of the adjacent land-
scapes. Accordingly, the role of carnivore dominance relationships AU T H O R C O N T R I B U T I O N S
and ecological similarity in mediating co-­occurrence patterns was Gonçalo Curveira-­Santos involved in conceptualization (lead); data
context-­specific. Such pair-­by-­area variability suggests that subor- curation (lead); formal analysis (lead); funding acquisition (support-
dinate carnivores may alternate between pre-­emptive behavioral ing); investigation (lead); methodology (lead); project administra-
strategies to avoid dominant competitors and instances of fine-­scale tion (equal); visualization (lead); writing—­original draft (lead); and
co-­occurrence (Karanth et al., 2017). The latter may or may not be writing—­review and editing (lead). Laura Gigliotti: involved in concep-
accompanied by strategies to mitigate competitive stress at shared tualization (supporting); formal analysis (supporting); investigation
sites, such as increased vigilance or facultative character displace- (supporting); methodology (supporting); writing—­original draft (sup-
ment (e.g., prey switching; Pfennig et al., 2006). The resulting co-­ porting); and writing—­review and editing (equal). Chris Sutherland
occurrence pattern and, importantly, the fitness impacts of species involved in conceptualization (supporting); formal analysis (sup-
co-­occurrence in a given context, likely depends on the complex in- porting); investigation (supporting); methodology (supporting); su-
terplay of density-­and trait-­mediated (e.g., phenotypic plasticity and pervision (equal); and writing—­review and editing (equal). Daniela
body-­size asymmetry) effects and characteristics of the local envi- Rato involved in conceptualization (supporting); formal analysis
ronment (e.g., landscape features, resource availability and diversity, (supporting); investigation (supporting); methodology (supporting);
and disturbance; Werner & Peacor, 2003). review and editing (supporting). Margarida Santos-­
 and writing—­
 By closely inspecting changes in co-­occurrence of species pairs Reis involved in conceptualization (supporting); funding acquisition
with spatial dependencies of varying signal across areas, we ob- (supporting); methodology (supporting); supervision (equal); and
served that spatial overlap between the same species was generally writing—­review and editing (equal). Lourens H. Swanepoel involved
more common in both the protected area and game ranches, and in conceptualization (supporting); data curation (supporting); formal
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 of 17 CURVEIRA-­SANTOS et al.

analysis (supporting); funding acquisition (lead); investigation (sup- Curveira-­Santos, G., Sutherland, C., Tenan, S., Fernández-­Chacón,
porting); methodology (supporting); project administration (lead); A., Mann, G. K. H., Pitman, R. T., & Swanepoel, L. H. (2021b).
 Mesocarnivore community structuring in the presence of Africa's
supervision (equal); and writing—­review and editing (equal).
 apex predator. Proceedings of the Royal Society B: Biological Sciences,
 288, 20202379.
AC K N OW L E D G M E N T S Curveira-­Santos, G., Gigliotti, L., Silva, A. P., Sutherland, C., Foord, S.,
We thank the iSimangaliso Authority, Ezemvelo KZN Wildlife, the Santos-­Reis, M., & Swanepoel, L. H. (2022). Broad aggressive inter-
 actions among African carnivores suggest intraguild killing is driven
Mun-­
 ya-­
 wana Conservancy, reserves' staff, and ranch-­
 owners
 by more than competition. Ecology, 103, 1–­12.
for supporting this research. We also thank everyone who as- Cusack, J. J., Dickman, A. J., Kalyahe, M., Rowcliffe, J. M., Carbon, C.,
sisted with fieldwork. This research was funded by South Africa's Macdonald, D. W., & Coulson, T. (2016). Revealing kleptoparasitic
National Research Foundation (UID: 107099 and 115040), African and predatory tendencies in an African mammal community using
 camera traps: a comparison of spatiotemporal approaches. Oikos,
Institute for Conservation Ecology, National Geographic Society
 125, 336–­3 42.
(EC-­314R-­18) and Wild Tomorrow Fund. G.C.-­S. and M.S.-­R . were Davis, C. L., Rich, L. N., Farris, Z. J., Kelly, M. J., Di Bitetti, M. S., Di
funded by Fundacão para a Ciência e a Tecnologia in the frame of Blanco, Y., Albanesi, S., Farhadinia, M. S., Gholikhani, N., Hamel,
a doctoral grant (PD/BD/114037/2015) and the research unit (UID/ S., Harmsen, B. J., Wultsch, C., Kane, M. D., Martins, Q., Murphy,
 A. J., Steenweg, R., Sunarto, S., Taktehrani, A., Thapa, K., … Miller,
BIA/00329/2019), respectively.
 D. A. W. (2018). Ecological correlates of the spatial co-­occurrence
 of sympatric mammalian carnivores worldwide. Ecology Letters, 21,
C O N FL I C T O F I N T E R E S T 1401–­1412.
The authors have no competing interests or conflict of interest to Di Bitetti, M. S., De Angelo, C. D., Di Blanco, Y. E., & Paviolo, A. (2010).
declare. Niche partitioning and species coexistence in a neotropical felid as-
 semblage. Acta Oecologica, 36, 403–­412.
 DiMiceli, C. M., Carroll, M. L., Sohlberg, R. A., Huang, C., Hansen, M.
DATA AVA I L A B I L I T Y S TAT E M E N T C., & Townshend, J. R. G. (2011). MODIS vegetation continuous
Data are available via the figshare repository https://doi. fields. Retrieved from: https://modis.gsfc.nasa.gov/data/datap​rod/
org/10.6084/m9.figsh​are.16965​340.v1. mod44.php
 Di Minin, E., MacMillan, D. C., Goodman, P. S., Escott, B., Slotow, R., &
 Moilanen, A. (2013). Conservation businesses and conservation
ORCID planning in a biological diversity hotspot. Conservation Biology, 27,
Gonçalo Curveira-­Santos https://orcid. 808–­820.
org/0000-0002-7136-5088 Donadio, E., & Buskirk, S. W. (2006). Diet, morphology, and interspecific
 killing in carnivora. American Naturalist, 167, 524–­536.
Laura Gigliotti https://orcid.org/0000-0002-6390-4133
 Dorazio, R. M., & Royle, J. A. (2005). Estimating size and composition
Chris Sutherland https://orcid.org/0000-0003-2073-1751 of biological communities by modeling the occurrence of species.
Margarida Santos-­Reis https://orcid.org/0000-0002-0337-963X Journal of the American Statistical Association, 100, 389–­398.
Lourens H. Swanepoel https://orcid.org/0000-0002-9955-8076 Dormann, C. F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré,
 G., Marquéz, J. R. G., Gruber, B., Lafourcade, B., Leitão, P. J.,
 Münkemüller, T., Mcclean, C., Osborne, P. E., Reineking, B.,
REFERENCES Schröder, B., Skidmore, A. K., Zurell, D., & Lautenbach, S. (2013).
Amarasekare, P. (2003). Competitive coexistence in spatially structured Collinearity: A review of methods to deal with it and a simulation
 environments: A synthesis. Ecology Letters, 6, 1109–­1122. study evaluating their performance. Ecography, 36, 27–­46.
Balme, G. A., Slotow, R., & Hunter, L. T. B. (2010). Edge effects and the Dorresteijn, I., Schultner, J., Nimmo, D. G., Fischer, J., Hanspach, J.,
 impact of non-­protected areas in carnivore conservation: Leopards Kuemmerle, T., Kehoe, L., & Ritchie, E. G. (2015). Incorporating
 in the Phinda-­Mkhuze complex, South Africa. Animal Conservation, anthropogenic effects into trophic ecology: Predator –­prey inter-
 13, 315–­323. actions in a human-­dominated landscape. Proceedings of the Royal
Bar-­Massada, A., & Belmaker, J. (2017). Non-­stationarity in the co-­ Society B: Biological Sciences, 282, 20151602.
 occurrence patterns of species across environmental gradients. Douma, J. C., & Weedon, J. T. (2019). Analysing continuous propor-
 Journal of Ecology, 105, 391–­399. tions in ecology and evolution: A practical introduction to beta
Blanchet, F. G., Cazelles, K., & Gravel, D. (2020). Co-­occurrence is not and Dirichlet regression. Methods in Ecology and Evolution, 10,
 evidence of ecological interactions. Ecology Letters, 23, 1050–­1063. 1412–­1430.
Caro, T. (2015). Conservation in the African Anthropocene. Protecting the Durant, S. M. (1998). Competition refuges and coexistence: An example
 wild (pp. 164–­169). Island Press/Center for Resource Economics. from Serengeti carnivores. Journal of Animal Ecology, 67, 370–­386.
Caro, T., & Stoner, C. (2003). The potential for interspecific competition Estes, J. A., Terborgh, J., Brashares, J. S., Power, M. E., Berger, J., Bond,
 among African carnivores. Biological Conservation, 110, 67–­75. W. J., Carpenter, S. R., Essington, T. E., Holt, R. D., Jackson, J. B.
Chamberlain, S. A., Bronstein, J. L., & Rudgers, J. A. (2014). How context C., Marquis, R. J., Oksanen, L., Oksanen, T., Paine, R. T., Pikitch,
 dependent are species interactions? Ecology Letters, 17, 881–­890. E. K., Ripple, W. J., Sandin, S. A., Scheffer, M., Schoener, T. W., …
Curveira-­Santos, G., Marques, T. A., Björklund, M., & Santos-­Reis, M. Wardle, D. A. (2011). Trophic downgrading of planet earth. Science,
 (2017). Mediterranean mesocarnivores in spatially structured 333, 301–­3 06.
 managed landscapes: Community organisation in time and space. Farris, Z. J., Gerber, B. D., Valenta, K., Rafaliarison, R.,
 Agriculture, Ecosystems & Environment, 237, 280–­289. Razafimahaimodison, J. C., Larney, E., Rajaonarivelo, T., Randriana,
Curveira-­Santos, G., Sutherland, C., Santos-­Reis, M., & Swanepoel, L. Z., Wright, P. C., & Chapman, C. A. (2017). Threats to a rainfor-
 H. (2021a). Responses of carnivore assemblages to decentralized est carnivore community: A multi-­year assessment of occupancy
 conservation approaches in a south African landscape. Journal of and co-­o ccurrence in Madagascar. Biological Conservation, 210,
 Applied Ecology, 58, 92–­103. 116–­124.
|

 20457758, 2022, 8, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ece3.9239 by Cochrane Portugal, Wiley Online Library on [09/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
CURVEIRA-­SANTOS et al. 13 of 17

Gaynor, K. M., Hojnowski, C. E., Carter, N. H., & Brashares, J. S. (2018). Monterroso, P., Díaz-­Ruíz, F., Lukacs, P. M., Alves, P. C., & Ferreras, P.
 The influence of human disturbance on wildlife nocturnality. (2020). Ecological traits and the spatial structure of competitive co-
 Science, 360, 1232–­1235. existence among carnivores. Ecology, 101, e03059.
Gelman, A., Carlin, J. B., Stern, H. S., & Rubin, D. (2013). Bayesian Data Mossaz, A., Buckley, R. C., & Castley, J. G. (2015). Ecotourism contri-
 Analysis. Chapman and Hall/CRC Press. butions to conservation of African big cats. Journal for Nature
Haswell, P. M., Kusak, J., & Hayward, M. W. (2015). Large carnivore im- Conservation, 28, 112–­118.
 pacts are context-­dependent. Food Webs, 12, 3–­13. Niedballa, J., Sollmann, R., Courtiol, A., & Wilting, A. (2016). cam-
Hayward, M. W., & Kerley, G. I. H. (2008). Prey preferences and dietary trapR: An R package for efficient camera trap data management.
 overlap amongst Africa's large predators. African Journal of Wildlife Methods in Ecology and Evolution, 7(12), 1457–­1462. https://doi.
 Research, 38, 93–­108. org/10.1111/2041-210X.12600
Hayward, M. W., & Slotow, R. (2009). Temporal partitioning of activity Oliveira-­Santos, L. G. R., Zucco, C. A., & Agostinelli, C. (2013). Using con-
 in large African carnivores: Tests of multiple hypotheses. South ditional circular kernel density functions to test hypotheses on ani-
 African Journal of Wildlife Research, 39, 109–­125. mal circadian activity. Animal Behaviour, 85, 269–­280.
Heim, N., Fisher, J. T., Volpe, J., Clevenger, A. P., & Paczkowski, J. (2019). Oriol-­cotterill, A., Valeix, M., Frank, L. G., Riginos, C., & Macdonald, D.
 Carnivore community response to anthropogenic landscape W. (2015). Landscapes of coexistence for terrestrial carnivores:
 change: Species-­specificity foils generalizations. Landscape Ecology, The ecological consequences of being downgraded from ultimate
 34, 2493–­2507. to penultimate predator by humans. Oikos, 124, 1263–­1273.
Hoeks, S., Huijbregts, M. A. J., Busana, M., Harfoot, M. B. J., Svenning, Packer, C., Loveridge, A., Canney, S., Caro, T., Garnett, S. T., Pfeifer, M.,
 J. C., & Santini, L. (2020). Mechanistic insights into the role of large Zander, K. K., Swanson, A., MacNulty, D., Balme, G., Bauer, H.,
 carnivores for ecosystem structure and functioning. Ecography, 43, Begg, C. M., Begg, K. S., Bhalla, S., Bissett, C., Bodasing, T., Brink,
 1752–­1763. H., Burger, A., Burton, A. C., … Polasky, S. (2013). Conserving large
Hui, F. K. C., Taskinen, S., Pledger, S., Foster, S. D., & Warton, D. I. (2015). carnivores: Dollars and fence. Ecology Letters, 16, 635–­6 41.
 Model-­based approaches to unconstrained ordination. Methods in Pfennig, D. W., Rice, A. M., & Martin, R. A. (2006). Ecological opportunity
 Ecology and Evolution, 6, 399–­411. and phenotypic plasticity interact to promote character displace-
Hunter, L. (2018). Carnivores of the world, 2nd ed. Princeton University ment and species coexistence. Ecology, 87, 769–­779.
 Press. Pitman, R. T., Fattebert, J., Williams, S. T., Williams, K. S., Hill, R. A.,
Jachowski, D. S., Butler, A., Eng, R. Y. Y., Gigliotti, L., Harris, S., & Williams, Hunter, L. T. B., Slotow, R., & Balme, G. A. (2017). The conservation
 A. (2020). Identifying mesopredator release in multi-­predator sys- costs of game ranching. Conservation Letters, 10, 403–­413.
 tems: a review of evidence from North America. Mammal Review, Plummer, M. (2003). JAGS: a program for analysis of Bayesian graphical
 50(4), 367–­381. models using Gibbs sampling. Proceedings of the 3rd International
Jonathan Davies, T., Meiri, S., Barraclough, T. G., & Gittleman, J. L. (2007). Workshop on Distributed Statistical Computing. https://www.r-proje​
 Species co-­existence and character divergence across carnivores. ct.org/confe​rence​s/DSC-2003/
 Ecology Letters, 10, 146–­152. Prugh, L. R., & Sivy, K. J. (2020). Enemies with benefits: Integrating posi-
Karanth, K. U., Srivathsa, A., Vasudev, D., Puri, M., Parameshwaran, R., tive and negative interactions among terrestrial carnivores. Ecology
 & Kumar, N. S. (2017). Spatio-­temporal interactions facilitate large Letters, 23, 902–­918.
 carnivore sympatry across a resource gradient. Proceedings of the Prugh, L. R., Stoner, C. J., Epps, C. W., Bean, W. T., Ripple, W. J., Laliberte,
 Royal Society B: Biological Sciences, 284, 20161860. A. S., & Brashares, J. S. (2009). The rise of the Mesopredator.
Kuijper, D. P. J., Sahlén, E., Elmhagen, B., Chamaillé-­Jammes, S., Sand, Bioscience, 59, 779–­791.
 H., Lone, K., & Cromsigt, J. P. G. M. (2016). Paws without claws? Ramesh, T., Kalle, R., & Downs, C. T. (2017). Staying safe from top pred-
 Ecological effects of large carnivores in anthropogenic land- ators: Patterns of co-­occurrence and inter-­predator interactions.
 scapes. Proceedings of the Royal Society B: Biological sciences, 283, Behavioral Ecology and Sociobiology, 71, 41.
 20161625. Rich, L. N., Miller, D. A. W., Robinson, H. S., McNutt, J. W., & Kelly, M. J.
Levi, T., & Wilmers, C. (2012). Wolves –­Coyotes –­Foxes: a cascade (2017). Carnivore distributions in Botswana are shaped by resource
 among carnivores. Ecology, 93, 921–­929. availability and intraguild species. Journal of Zoology, 303, 90–­98.
Lindsey, P. A., Romaach, S. S., & Davies-­Mostert, H. T. (2008). A syn- Ridout, M. S., & Linkie, M. (2009). Estimating overlap of daily activity
 thesis of early indicators of the drivers of predator conserva- patterns from camera trap data. Journal of Agricultural, Biological,
 tion on private lands in South Africa. In M. W. Hayward & M. J. and Environmental Statistics, 14, 322–­337.
 Somers (Eds.), Reintroduction of top-­order predators (pp. 321–­3 44). Ripple, W. J., Estes, J. A., Beschta, R. L., Wilmers, C. C., Ritchie, E. G.,
 Wiley-­Blackwell. Hebblewhite, M., Berger, J., Elmhagen, B., Letnic, M., Nelson, M. P.,
Macarthur, R., & Levins, R. (1967). The limiting similarity, convergence, Schmitz, O. J., Smith, D. W., Wallach, A. D., & Wirsing, A. J. (2014).
 and divergence of coexisting species. The American Naturalist, 101, Status and ecological effects of the world's largest carnivores.
 377–­385. Science, 343, 1241484.
Miller, J. R. B., Pitman, R. T., Mann, G. K. H., Fuller, A. K., & Balme, G. Ritchie, E. G., & Johnson, C. N. (2009). Predator interactions, meso-
 A. (2018). Lions and leopards coexist without spatial, temporal or predator release and biodiversity conservation. Ecology Letters, 12,
 demographic effects of interspecific competition. Journal of Animal 982–­998.
 Ecology, 87, 1709–­1726. Ritchie, E. G., Elmhagen, B., Glen, A. S., Letnic, M., Ludwig, G., &
Mills, K. L., & Harris, N. C. (2020). Humans disrupt access to prey for large McDonald, R. A. (2012). Ecosystem restoration with teeth: What
 African carnivores. eLife, 9, 1–­16. role for predators? Trends in Ecology and Evolution, 27, 265–­271.
Mills, D. R., San, E. D. L., Robinson, H., Isoke, S., Slotow, R., & Hunter, L. Robinson, Q. H., Bustos, D., & Roemer, G. W. (2014). The application of
 (2019). Competition and specialization in an African forest carni- occupancy modeling to evaluate intraguild predation in a model
 vore community. Ecology and Evolution, 9, 10092–­10108. carnivore system. Ecology, 95, 3112–­3123.
Monterroso, P., Alves, P. C., & Ferreras, P. (2014). Plasticity in cir- Rollinson, C. R., Finley, A. O., Alexander, M. R., Banerjee, S., Dixon Hamil,
 cadian activity patterns of mesocarnivores in southwestern K.-­A ., Koenig, L. E., Locke, D. H., DeMarche, M. L., Tingley, M. W.,
 Europe: Implications for species coexistence. Behavioral Ecology Wheeler, K., Youngflesh, C., & Zipkin, E. F. (2021). Working across
 and Sociobiology, 68, 1403–­1417. https://doi.org/10.1007/s0026​ space and time: Nonstationarity in ecological research and applica-
 5-­014-­1748-­1 tion. Frontiers in Ecology and the Environment, 19, 66–­72.
You can also read